Volume, Gain, and Loudness: Making Audio Sound Just Right

Why doesn't pushing the volume to 200% sound much louder — just harsher and distorted? dB, dBFS, LUFS: what's the difference between these three "decibels"?

You've almost certainly run into situations like these: the same video is barely audible through your phone's speaker but deafening on your TV; two songs play back-to-back in your music app, one gentle and one blasting, forcing you to keep reaching for the volume control; you push a voice memo to 200% volume and, sure, it's louder—but now it's accompanied by harsh, crackling distortion.

All of these frustrations share a common root: what everyday language lumps together as "volume" is actually several distinct concepts in audio engineering—gain, level, peak, and loudness—each governed by its own measurement rules. Without understanding those rules, adjusting volume will always be guesswork. This article explains them all, once and for all.

The Decibel (dB): A Logarithmic Ruler

Almost everything about audio level is measured in decibels (dB), and the most commonly misunderstood thing about decibels is this: the scale is logarithmic, not linear.

For sound amplitude, the decibel is defined as:

dB = 20 × log10(amplitude ratio)

This yields a few correspondences worth memorizing:

Here's the counterintuitive part: doubling the amplitude (+6 dB) does not make something sound twice as loud. Human loudness perception is itself roughly logarithmic, and psychoacoustic research suggests that perceived loudness doubles at around +10 dB. That's why dragging a volume slider from 50% to 100% (+6 dB) sounds like "a bit louder," not "twice as loud."

Also keep in mind: a decibel value is always relative. "60 dB" on its own is meaningless—you have to state the reference. That need gives rise to a whole family of decibel scales for different contexts.

dBFS: The Ceiling of Digital Audio

Inside the digital domain—audio files, editing software, and processing tools—the relevant scale is dBFS (decibels relative to Full Scale):

• 0 dBFS is the maximum amplitude a digital system can represent—the ceiling;
• All real-world signals sit at negative dBFS values, such as -6 dBFS or -20 dBFS;
• The closer to 0, the "hotter" the signal; very negative values (like -40 dBFS) mean the sound is faint.

   0 dBFS ──────── the ceiling (clipping line)
  -6 dBFS ──────── amplitude is half of maximum
 -12 dBFS ──────── amplitude is a quarter of maximum
  -∞ dBFS ──────── silence

Clipping: Irreversible Distortion

When a signal is pushed past the 0 dBFS ceiling, the excess is simply sliced off—this is clipping. Clipping produces a harsh, buzzing distortion, and once it happens, the damage is permanent: the flattened waveform can't be restored by "turning it back down" later, because the information beyond the ceiling is gone forever.

That's why professional workflows live by an iron rule: record too quiet rather than too hot. A quiet recording can be gained up later with no penalty (within reason); a clipped one is damaged goods for good.

Gain vs. Volume: Are They the Same Thing?

The two words are used interchangeably in everyday speech, but audio engineering draws a subtle, important distinction:

• Gain: amplification or attenuation applied to the signal itself—it changes the audio data. The "volume" parameter in audio tools is, in essence, gain.
• Volume: usually refers to output-level control at the end of the playback chain (system volume, player volume). It leaves the file untouched.

In other words, gain modifies the file; volume modifies playback. When you apply +6 dB of gain in a tool and export, the file itself becomes louder. When you turn up your player's volume, the file doesn't change at all.

Converting Percentages to Decibels

Many tools express gain as a percentage (100% = original level). A percentage is just another way of stating a decibel change:

Be careful with any gain above 100%: if the original audio already peaks close to 0 dBFS, boosting it guarantees clipping. For example, audio peaking at -3 dBFS can only be safely boosted by about +3 dB (roughly 140%).

Same Peaks, Wildly Different Loudness—Why?

Everything so far has been about peaks—the maximum amplitude a signal reaches at any instant. But peaks are not what your ears perceive as "loud."

Imagine two audio clips: one is steady, continuous speech; the other is mostly silence with a few drum hits. Their peaks might be identical (both the loudest drum hit and the loudest syllable touch -3 dBFS), yet the speech clearly sounds "louder." The reasons:

• Human hearing responds to the energy of sound averaged over time, not to instantaneous peaks;
• Frequency also shapes loudness perception—the ear is most sensitive to midrange frequencies around 2–5 kHz and relatively deaf to very low and very high frequencies (the equal-loudness contours).

So describing "how loud" something is using peaks alone doesn't work. The industry needed a measurement that genuinely tracks human perception.

Loudness and LUFS: A Perception-Aligned Measurement

Loudness is the concept invented to solve exactly this problem, and its modern standard unit is the LUFS (Loudness Units relative to Full Scale; essentially equivalent to LKFS). Defined by the ITU-R BS.1770 standard, it works by:

1. Applying K-weighting filtering to the audio, mimicking how ear sensitivity varies across frequencies;
2. Integrating energy over time, with a gating mechanism that excludes very quiet passages (so long silences don't drag down the average).

Like dBFS, LUFS tops out at 0 with negative values below—but it measures perceived loudness, not peak amplitude. Two intuitions to internalize:

• If program A measures -14 LUFS and program B measures -20 LUFS, A sounds roughly 6 loudness units louder than B (1 LU ≈ 1 dB of loudness change);
• Audio with high peaks but sparse content tends to measure low in LUFS, while audio with modest peaks but dense, sustained content measures high.

Two companion metrics are worth knowing:

• True Peak (dBTP): accounts for inter-sample peaks that can occur when digital audio is converted back to analog—stricter than sample peak. Platforms and mastering engineers typically require true peaks no higher than -1 dBTP.
• Loudness Range (LRA): describes how much loudness varies within a program. Films have a large LRA (hushed dialogue to explosions); ads and pop music have a small one (loud throughout).

Loudness Normalization: What Platforms Do Behind Your Back

Loudness normalization means adjusting audio to a uniform target loudness. It's fundamentally different from peak normalization (raising every file's peak to the same level), which does nothing to fix inconsistent perceived loudness.

Loudness targets used by major platforms and standards (for reference—check official documentation for current values):

This has an important practical consequence: mastering your music or video far louder than the platform's target is pointless—the platform will simply turn it down, and the dynamic range you sacrificed to get "extra loud" was spent for nothing. The correct modern approach is to master close to the platform's target loudness and leave 1 dB of true-peak headroom.

Dynamic Range: The Space Between Loud and Quiet

Dynamic range is the gap, in decibels, between the loudest and quietest parts of an audio program:

• A live symphony: dynamic range can reach 40–60 dB, from barely-there pianissimo to a wall of sound;
• A finished pop track: often compressed down to 6–10 dB, staying loud from start to finish;
• A spoken podcast: typically 10–20 dB, keeping speech intelligible in any listening environment.

One common confusion deserves special attention: dynamic range compression has nothing to do with file compression (like MP3). The former is an audio effect (a compressor) that narrows volume swings; the latter is an encoding technique that shrinks file size. They share a word and nothing else.

From the late 1990s through the 2000s, the music industry fought the infamous "loudness war"—labels crushed dynamic range to squeeze out every last decibel, hoping to overpower competing tracks on the radio, at the cost of musical light and shade. The spread of streaming loudness normalization is finally ending that war.

Common Misconceptions

1. "Cranking volume to maximum never hurts." Anything past 0 dBFS clips, and clipping is irreversible. Check peak headroom before boosting gain.
2. "Decibels and percentages map directly onto how loud something feels." +6 dB doubles the amplitude, but the sound is far from "twice as loud" (that takes roughly +10 dB).
3. "Matching peaks means matching loudness." Peak normalization doesn't do the job—loudness matching requires LUFS.
4. "Re-boosting an MP3's gain repeatedly is harmless." Every re-encode of a lossy format compounds quality loss. Always work from the original file in a single pass.
5. "Louder is better." Platform normalization erases any loudness advantage, while the dynamic-range damage from over-compression is yours to keep.

Practical Tips

• Leave headroom: keep peaks below -1 dBTP when producing and exporting, leaving a safety margin for encoding and playback chains.
• Master to the target platform: around -14 LUFS for video soundtracks, around -16 LUFS for podcasts. Don't chase loudness for its own sake.
• Adjust gain instead of re-encoding repeatedly: volume adjustment is a gain operation—do it once, from the original file, to avoid generational loss in lossy formats.
• Measure peaks before boosting: your safe gain ceiling is 0 dBFS minus the current peak (leaving 1 dB extra is wiser).
• Test on different devices: phone speakers, headphones, and car stereos have wildly different frequency responses. Verify important content on at least two.
• For speech, prioritize clarity: moderate dynamic range compression (so quiet passages remain audible) is more effective than simply turning everything up.

Further Reading

• Audio Cutting and Merging: Trimming and Joining Sound — zero crossings, fades, and choosing audio formats
• Audio Speed and Pitch: The Art of Time-Stretching — what happens to sound when you change its speed

If you just need to quickly adjust the volume of an audio file, you can use this site's volume adjustment tool—and now you know that behind that "volume percentage" slider lies the logarithmic world of gain and decibels, where 150% is really just a gentle +3.5 dB lift.

Related Tools

• Audio Volume
• Add Audio